← ClaudeAtlas

app-blueprintlisted

Use this skill when a user wants to start building a new app, product, or feature and the requirements are not yet clear — phrases like "I want to build an app", "help me create a project", "let's start a new app", "I have an idea for a tool", or when a user jumps straight to asking for code/architecture for a project that has no defined scope yet. It runs a short, plain-language interview to pin down the essential product decisions BEFORE any code or architecture is proposed, so the assistant never silently guesses scope, data model, users, or trade-offs on the user's behalf. Skip this skill if the user already provided a clear written spec, or explicitly asks to skip discovery.
PostDev360/postdev-skills · ★ 1 · AI & Automation · score 55
Install: claude install-skill PostDev360/postdev-skills
# App Blueprint ## Purpose Non-technical founders and early-stage builders often ask an AI assistant to "build my app" without realizing how many architectural decisions get made — silently — on their behalf along the way. This skill forces those decisions into the open first, as a short conversation in plain language, so the person making the product owns the trade-offs instead of the AI guessing them. The output is a written brief the user can read and correct, which then becomes the foundation for any architecture or implementation work. ## Core rule **Never let a technical decision get made by default.** If a question below would normally be answered by a developer's judgment call (e.g. "does this need user accounts?", "should data survive a restart?", "one platform or several?"), it must be answered by the user in this interview — not inferred silently, even when the "obvious" answer seems obvious. If the user's answer to a question doesn't resolve a downstream decision, ask a follow-up. Do not fill the gap with an assumption. ## Communication style Keep every report, status update, and summary short and to the point, both during the interview and throughout any implementation work that follows the confirmed brief. Skip restating context the user already knows, and avoid padding explanations — this keeps token usage low across a long build. Brevity applies to the assistant's own commentary, not to the substance of the questions asked or the brief itself. Run the